Building iOS Apps From Scratch by Mike Rundle

Before taking a crack at any Design Then Code project tutorials you'll need some knowledge of Xcode, Objective-C, Cocoa and UIKit. My goal is for this guide to help bridge the gap between having no knowledge of iOS development and having enough to start tackling more interesting projects.

Tools

Apple provides a number of tools to enable developers to build Mac and iOS apps. To download them, head to the Mac App Store and search for "Xcode". This $4.99 download will give you access to Xcode (the IDE that Mac/iPhone developers use), Interface Builder, the Cocoa Frameworks, tools for testing your apps, and a lot more. To register as an official iOS developer and publish apps to the App Store (not to mention testing your apps on a real device!) it costs $99 per year.

Here's a quick overview of the tools Apple provides.

Xcode

Xcode is an IDE (Integrated Development Environment) used by Mac and iOS developers to build applications. It's not just a code editor: it has a variety of additional goodies baked in like great autocomplete support, static code analysis (it finds bugs in your code before you compile, including memory leaks) and a variety of debugging and performance tools. You could use TextMate or BBEdit and then use command line tools to do the final compilation, but most developers choose to do it all within Xcode. I use Xcode for all app development.

Interface Builder

Interface Builder is an application that lets you build your interfaces visually. Built-in objects like buttons, tab bars, sliders and labels can easily be dragged onto your app's interface and then configured by tweaking the palettes and panels. You can also use Interface Builder to connect targets and actions (what happens when an interface element is acted on by a user, and what object handles the action) as well as manipulate controllers and object bindings.

In my particular development workflow, I prefer not to use Interface Builder, mostly because I work on custom interface components and those still take a lot of code to get exactly right. I recommend that new Mac and iPhone developers get acquainted with Interface Builder, but at the same time still learn the UIKit code that it is generating for you. I've seen many developers start using Interface Builder and never leave it, so they never actually learn how to code interfaces from scratch. All Design Then Code tutorials forego Interface Builder and explain how to write all UI code by hand.

Frameworks

And the most important piece of the puzzle: frameworks. Without frameworks and APIs developers wouldn't easily be able to create applications that run on Mac OS X or iOS. Apple provides dozens of frameworks that enable developers to do things like create user interfaces, write networking code, encrypt important information, draw graphics to the screen, play audio and video, save data and passwords, take pictures, display webpages and much more.

The frameworks that Apple provides let you start off with a rich set of commands and tools upon which to build your applications. Without the various frameworks that Apple provides, every developer would be reinventing the wheel over and over again. There are lots of goodies given to you in the various Cocoa frameworks so, many times, an incredibly complex thing can be accomplished in only a few lines of code. An example of this is automatically fetching a file on the Internet by its URL, parsing this file, then stuffing it into a data structure that you can manipulate instantly is just a one-liner!.

These are just the main tools that Apple provides but there are many, many more to explore and use. Go pick up Xcode from the Mac App Store and start poking around within your new /Developer directory.

Introduction To Programming?

Are you totally new to computer programming? Have never even written any JavaScript? This tutorial might be tough to swallow. Take a look at the following list of terms and examples:

If any of these are confusing to you, I'd suggest heading to the JavaScript tutorial at w3schools.com and working through the first few sets of chapters to get a quick feel for general computer programming constructs. If you're coming from a language that does not look like C (Lisp, Ruby, etc.) or if you have no C programming knowledge (what's a header file? what's a struct?), I'd recommend taking a quick read of Scott Stevenson's excellent C language tutorial and then coming back after your brain is thoroughly soaked with knowledge.

Now let's talk about Objective-C.

Introduction To Objective-C

Objective-C is the language that most Mac and iOS developers use to write native applications. There are alternative ways to write Mac and iOS apps (pure C or C++, MacRuby, PyObjC, MonoTouch, etc.) but we won't be exploring those avenues here. Apart from these programming languages, it's also possible to build "apps" that aren't natively compiled for the platform but are instead websites made to look like a native app but loaded in a web browser. Those apps can be written using a variety of frameworks but they're primarily built using web technologies including JavaScript, HTML and CSS.

If you're looking to build mobile websites or web apps then you don't need to learn Objective-C or Cocoa and probably don't need to go any farther in this tutorial. If you're looking to learn a new programming language and build native apps for the Mac and iOS, then this is the place for you!

Objective-C is an object-oriented programming language that is essentially a thin layer on top of C. It adds Smalltalk-style messaging, runtime reflection, inheritance and many other things to the C programming language. Its syntax is very unique and may be confusing at first to seasoned developers used to languages like Ruby or Java. Because Objective-C is a superset of C, developers can "drop down into" C at any point in their code if they wish. Many parts of a Mac or iOS application will utilize native C function calls and primitive C data types right next Objective-C methods and data types.

Classes, Objects & Methods

In case your object-oriented programming fu is rusty (or totally nonexistent) we should first define what classes, objects and methods are and how they relate to each other in Objective-C.

A class is a blueprint that describes the state and behavior of a particular set of objects. It typically models a concept or a thing from the real world, for example, an Animal. An Animal class would probably define some variables like the number of legs it has, its height and weight, and also behaviors, like running, eating and sleeping. An object of a given class is called an instance of the class. Each new instance you create can have its own values for the data. You could create 100 instances of the Animal class and set each one's variables to something different. Methods are behaviors a class possesses, and in Objective-C, you can call a method directly on the class itself or on an instance of it. These two types of methods are called class methods and instance methods.

When you want to create a new instance of a class, you have to first allocate and initialize a block of memory for it on the heap. The heap is memory set aside for dynamic allocation and in Objective-C all objects live on the heap. All Obj-C objects are pointers to this block of memory and an asterisk (*) denotes that a variable is a pointer type. Here's an example of creating an instance of our Animal class.

Animal *myAnimal = [[Animal alloc] init];

Let's break this down piece by piece. The first thing you'll probably notice are the square brackets. They're everywhere in Objective-C and they encapsulate method calls. On the left side of the line we're creating a new Animal* variable named myAnimal, then, on the right side, we're doing 2 different method calls (one nested within the other) and assigning what they return to myAnimal.

To understand what's going on we need to look at the method calls from the inside-out. First, we're calling [Animal alloc] which means we're calling the class method +alloc directly on the Animal class. Class methods are preceded with a plus sign. This method returns a generic object type of id. Then, we call the instance method -init which initializes the memory, allowing us to actually use this new object. Instance methods start with a minus sign. Objects in Objective-C aren't fully ready to be used unless these two steps are taken.

An important concept to understand when working with objects is that if you allocated memory for it, you're also responsible for letting the runtime know when you're done using the object so that its memory can be released. An object can be in use in different parts of an app at once, so Objective-C uses a reference counting system to keep track of all this usage.

When an object is first created, the number of references on the object is one. When an object is no longer being used, a -release message must be sent to it, and this decreases the number of references on the object by one. When the overall number of references on an object drops to zero, the object's memory is freed. If objects that are no longer being used stay around and occupy memory some bad things could happen, one of which is your iOS app is forced to quit because it's using too many resources. For Mac apps, if an app takes up more and more memory, it could make the whole system sluggish. Have you ever left Photoshop or Safari open for a few days and everything gets bogged down? It's because the apps are continually using more memory — potentially due to a memory leak — and there are fewer resources for other running apps. Memory is a precious resource when developing software so it's important to use it judiciously. For an in-depth look at Cocoa memory management rules, read Apple's Memory Management Programming Guide.

Now that we know how to create a fully-formed, fully-functional instance of a class, let's create a new Airplane object and call some methods on it.

Airplane *myAirplane = [[Airplane alloc] init];
[myAirplane fly];
[myAirplane flyTo:@"Austin, TX"];
[myAirplane flyTo:@"Dulles Airport" landAtTerminal:3];

Calling methods on objects in Objective-C has a fairly unique syntax compared to other languages you may be familiar with, so let's recreate the same calls in Java.

Airplane myAirplane = new Airplane();
myAirplane.fly();
myAirplane.flyTo("Austin, TX");
myAirplane.flyToAndLandAtTerminal("Dulles Airport", 3);

Here we're initializing a new Airplane object and then calling 3 different methods on it. The first, -fly takes in no arguments. The second, -flyTo: takes in one argument. The third, -flyTo:landAtTerminal: takes in two arguments. The full names of Objective-C methods include the names of the arguments. Colons in a method name indicate that it takes an argument, one colon for each argument that it takes. These are instance methods so a minus sign is included at the beginning of the method name.

Objects don't just have behaviors, they can store data as well. Let's imagine a Car object and the types of data attributes it could possess:

Let's look at these attributes. First, we have model year. This could be represented by a numerical year, like 2008. What kind of data is 2008? It's a number, and in Objective-C that number could be represented in a few different ways:

Choices, choices! Let's talk about what each of these mean. The variable type int comes from C and is a primitive data type that holds a number with a certain maximum value. An NSInteger is a special primitive data type from Apple's Foundation framework that is automatically sized correctly for the current architecture. The third way is an instance of the NSNumber class that is also defined in Foundation framework. We'll learn more about Foundation and other Apple frameworks in a bit.

Neither an int nor an NSInteger are objects which means you don't have to worry about dynamically allocating memory for them since they have a pre-defined size. They get created on the stack, not the heap, so no pointer is needed to access their contents. Things like an int, NSInteger, CGFloat, CGPoint, CGRect (and a whole slew of other things defined in the Foundation Data Types Reference) aren't objects, so no pointer (asterisk) is needed.

Now let's look at the third example of how we could store the number 2008: as an NSNumber object.

We can initialize a new NSNumber object like this:

NSNumber *year = [NSNumber numberWithInt:2008];

If this is an instance of the NSNumber class, where's +alloc? Where's -init? Well, it turns out, some objects have convenient class methods that return an object with memory that you don't have to manage yourself. Since you didn't manually create the memory using +alloc and -init, you don't have to worry about sending it a -release message when you're done. Many frequently-used objects defined in Apple's Foundation framework have these nice constructors. Of course if you want (or need) to manually manage their memory, we could have also done this:

NSNumber *year = [[NSNumber alloc] initWithInt:2008];

So now we have an instance of the NSNumber class that is wrapped around a regular int of 2008. Why go through the trouble of using an NSNumber object when we could have more easily used an int or NSInteger? Because some Cocoa classes only make use of other objects, they simply don't work with primitive types, for example, NSArray. An NSArray is an object that manages the ordered collection of other objects. If we wanted to use an NSArray to hold a series of numbers, we'd have to use NSNumber objects because it doesn't work with primitive data types like int or NSInteger. We could use ints and NSIntegers in a regular C array, but then we wouldn't get all the great, built-in behaviors that are defined in the NSArray class.

To initialize our Car object and access its instance variables, we do this:

Car *myCar = [[Car alloc] init];
myCar.modelYear = [NSNumber numberWithInt:2008];

In this example, modelYear is a @property of the Car class which enables us to use the dot syntax to access it. We'll get into the specifics of properties and instance variables a bit later.

Now that we know how to initialize new objects, call methods, and access its instance variables, let's dive into defining new classes.

Defining An Objective-C Class

To define a new class in Objective-C you must define two things: the class's interface and its implementation. A class interface tells the compiler about the class's instance variables and methods so it know what to expect. Like a preview of upcoming attractions. A class implementation is the actual code and functionality for each method. By convention, a class interface goes into a .h file and the implementation is in a .m file. For a Car class, the two files would be Car.h and Car.m.

Here's our Car interface file:

@interface Car : NSObject {
	NSNumber *modelYear;
	NSString *manufacturerName;
	UIColor *color;
	BOOL willTurnOn;
}

@property (nonatomic, retain) NSNumber *modelYear;
@property (nonatomic, copy) NSString *manufacturerName;
@property (nonatomic, retain) UIColor *color;
@property (nonatomic, assign) BOOL willTurnOn;

- (void)drive;
- (void)turnRadioToStation:(NSString *)station;
- (void)setRadioVolume:(NSNumber *)volume;
- (BOOL)hasNavigation;

@end

Let's go over the Car class interface line by line.

@interface Car : NSObject

At the top of our interface file we declare Car to be a subclass of NSObject which is the root class for most of the Cocoa classes you'll encounter. I like to think about subclassing as creating a more specialized version of an object. A generic class in Cocoa should be an NSObject subclass. If you're making a specialized version of a button you might make a UIButton subclass. Specialized version of a text label? It'd be a UILabel subclass. What if we wanted to make a specialized version of a Car, say, a convertible? Well, we'd probably declare a Convertible class which would be a subclass of our Car class, which, as I just explained, is itself an NSObject subclass.

A subclass gives you the functionality of its parent class and all the other parent classes up the chain. A Ferrari object is a also a Convertible object which is also a Car object. If you declare all Car objects as having GPS then the Ferrari will as well. Right? Sure, but when you create a subclass in Objective-C you can choose to add or re-implement functionality of the parent classes as you see fit. For example, all UIButtons look a certain way, but if you want to make a totally custom button for your iOS app, you can subclass UIButton and draw the button's graphics yourself. Your custom code will override the default behavior in the parent UIButton class.

Next up, the class's instance variables.

@interface Car : NSObject {
	NSNumber *modelYear;
	NSString *manufacturerName;
	UIColor *color;
	BOOL willTurnOn;
}

Inside the curly braces, right after the class declaration, you'll find the object's instance variables. Instance variables are attributes for a specific instance of an object. If you create 3 different Car objects they would each have their own values for these instance variables. Not all cars have the same color, and not all instances of our Car class will have the same color instance variable value. Instance variables are declared with the variable type and the variable's name. Three of these instance variables are objects (see the asterisks?) and the fourth is a primitive boolean type that can only be either YES or NO.

After the instance variables you'll find a series of @property declarations.

@property (nonatomic, retain) NSNumber *modelYear;
@property (nonatomic, copy) NSString *manufacturerName;
@property (nonatomic, retain) UIColor *color;
@property (nonatomic, assign) BOOL willTurnOn;

Properties are an Objective-C 2.0 feature that allow you quicker, more effortless access to your instance variables. What these actually do is configure the getter and setter methods to access and update these instance variables. For each instance variable you want to access and update using dot syntax (car.color, car.willTurnOn, etc.), you have to declare it as a @property as well. Before Obj-C properties came along, you'd have to write two methods for each instance variable to "get" and "set" its value, but now you can use properties to configure them automatically, allowing us to use the nifty dot syntax described previously. There are various ways to configure @properties and rather than go into depth here, you'll find Apple has a guide just for them.

And finally you'll see the method signatures for the behaviors implemented in the .m file.

- (void)drive;
- (void)turnRadioToStation:(NSString *)station;
- (void)setRadioVolume:(NSNumber *)volume;
- (BOOL)hasNavigation;

These method declarations describe the type of data returned by a method, the method name and its arguments. These are all instance methods because of the - preceding each one. These must match their corresponding implementation (the code that actually runs when the method is called) in the .m file exactly.

Now for the implementation of our Car class which will exist in a file called Car.m.

@implementation Car

@synthesize modelYear, manufacturerName, color, willTurnOn;

- (void)drive {
	// Code to make the car drive would go here!
}

- (void)turnRadioToStation:(NSString *)station {
	// Turn on the radio adjust the station!
}

- (void)setRadioVolume:(NSNumber *)volume {
	// This one goes to 11!
}

- (BOOL)hasNavigation {
	// Does this car have it?
}

@end

Again, let's go over each line.

@implementation Car

At the top is a compiler directive of @implementation which signifies the start of our implementation file for this class.

@synthesize modelYear, manufacturerName, color, willTurnOn;

Next is the @synthesize statement which is the counterpart to @property in the class interface. In our interface we declared our object's properties, and now, in the implementation file, we use @synthesize to automatically generate the getters and setters based on our declared configurations for them. Don't forget to have both @property and @synthesize declarations for each property you want to use.

- (void)drive {
	// Code to make the car drive would go here!
}

- (void)turnRadioToStation:(NSString *)station {
	// Turn on the radio adjust the station!
}

- (void)setRadioVolume:(NSNumber *)volume {
	// This one goes to 11!
}

- (BOOL)hasNavigation {
	// Does this car have it?
}

After we've synthesized our properties, we will write the code for each of the methods declared in the interface. The method signature is the same as in the interface file, but the functionality for each method follows directly after within a pair of curly braces.

And that's it! Our Car class is now fully defined. Let's initialize a new Car object and work with it a bit.

Car *myCar = [[Car alloc] init];

// Set some of its properties
myCar.color = [UIColor redColor];
myCar.modelYear = [NSNumber numberWithInt:1995];

// Call some of its methods
[myCar turnRadioToStation:@"Hot 97"];
[myCar drive];

// Tell the Objective-C runtime that we're done with this Car object
[myCar release];

After creating a fresh instance of a Car object we can start using it. First, we can set some of its properties like its color or modelYear. The dot notation "object.property" works because we declared certain instance variables to be able to be retrieved or set in this manner. It doesn't happen automatically, but if you use @property and @synthesize then you're golden. It may seem pretty simple, but Objective-C is doing some legwork in the background to make sure that memory is allocated and deallocated properly when you directly access an object's properties.

We can also call a Car object's instance methods like -turnRadioToStation: and -drive. Notice in the first method call, we're passing in a string as an argument since that is expected from the method definition. The second method, -drive, takes in no arguments.

When we're done using our instance of a Car class, we send it a -release message to decrease its reference count by one. As the only place using this object, the reference count would drop to zero and its memory would be automatically deallocated. To be more specific, the runtime automatically calls the -dealloc method on our Car object. Typically we would implement -dealloc in this class and do some cleanup/maintenance work before the memory is deallocated and the object vanishes.

This is just scratching the surface of Objective-C, but it's enough to propel you in the right direction if you want to start understanding and building Mac and iOS apps.

Now onto the interesting stuff.

Introduction To Cocoa (and Cocoa Touch)

The phrases "Objective-C development" and "Cocoa development" are thrown around interchangeably, and usually if you say either one, people know that you're talking about building Mac and iOS apps.

Cocoa is a set of frameworks that Apple built for developers to use when building Mac and iOS apps. The Cocoa frameworks are written in Objective-C and Objective-C is the preferred language used to access the Cocoa frameworks when developing Mac and iOS applications. Apple used to provide support for using Java to access the Cocoa frameworks, but it has fallen out of use in recent years. There are other ways to access Cocoa APIs, including bindings for Python, Perl, Ruby, and C#, but for the most part those aren't sanctioned by Apple.

The frameworks that make up what's known as "Cocoa" include Foundation and Application Kit (AppKit) for Mac OS X, and when developing iOS apps, Foundation and UIKit. Foundation framework provides a base layer of classes for string and number manipulation, date objects, collections, networking and more. Some objects include NSString, NSArray, NSURLConnection, NSDictionary and NSNumber. AppKit classes are used to build the interface for Mac applications including windows, controls and menus. Example AppKit objects include NSWindow, NSView, NSButton, NSImageView and NSScrollView.

When developing iOS apps you'll still use the same, more generic classes available in the Foundation framework but instead of using AppKit to build user interfaces, UIKit is used instead. Some objects from UIKit that you'll be using all the time are UIWindow, UIView, UIButton, UIViewController, UILabel, UITableView, UITextView and UIWebView. You might notice that for many objects, the UIKit version has the same name as the AppKit original, but with a UI- prefix instead of NS-. A full list of classes available within UIKit is shown here at Apple's Developer site.

To be honest, you'll probably never use Car or Airplane objects unless you're building an app for a car dealership or airport. The objects that you'll typically be using and subclassing are ones that Apple provides as part of the Cocoa frameworks, like NSArray to manage collections of objects, UIViewController to manage a screenful of content, UITableViewCell for a custom row in a table, UILabel for displaying small bits of text, UIImageView to display an image on the screen and many, many more. Apple provides many interesting and incredibly useful classes for you to use as you construct Mac and iOS apps, and typically they are descendants of a number of other classes, adding more functionality and specialization the further down the inheritance chain you look. For example, let's take a look at UIButton's full inheritance chain:

NSObject is the root class for most Cocoa objects and handles memory allocation, initialization, message passing and other lower-level tasks. UIResponder handles interaction events like tapping the screen with your finger. UIView defines a rectangular area on the screen and provides the functionality to place or draw content within that rectangle. UIControl is the base class for user interface widgets and manages the state (selected? highlighted? disabled?) as well as what actions occur when the user interacts with it. And finally we get down to UIButton, a specialized control that takes care of a button's text label, all the styling of the button, and what it looks like when it's selected or highlighted.

It's important to remember that when using a UIButton you don't just have access to the functionality of that class, but of all the classes that UIButton inherits from all the way up to NSObject. I can't tell you how many times I missed something that was available in an object's parent class because I was only looking in the documentation for the immediate class I was using.

Common Objects & Functions

Regardless of what your app looks like there are some objects, functions and data structures that you'll use over and over so it's important to become familiar with them.

NSString

A string is a sequence of characters, and in Objective-C, a string is an NSString object with a lot of built-in functionality. The shorthand way of referring to a string @"looks like this" with an @ sign in front of a double-quoted run of characters. NSString objects can be compared to one another, converted to other encoding formats, used to hold the contents of a file or URL, changed to C-style strings, used in regular expression matches or turned into numbers.

Once you create a string it cannot be modified. This may sound odd if you're coming from PHP or JavaScript, but if you want to modify a string after it has been created, you should use NSMutableString instead, a mutable (modifiable) subclass of NSString.

Here's an example of how you might construct an NSURL object by using a string:

NSURL *url = [NSURL URLWithString:@"http://google.com/"];

And here's an example of using a mutable string:

NSMutableString *myString = [NSMutableString stringWithString:@"Reading"];
[myString replaceOccurrencesOfString:@"Read" withString:@"Writ"
  options:NSLiteralSearch range:NSMakeRange(0, [myString length])];

That's a gigantic method that we're calling! The NSMutableString instance method -replaceOccurrencesOfString:
withString:options:range:
finds a string within a string, then replaces it with something else. Notice the call to the NSMakeRange() function, part of the Foundation framework. It returns an NSRange struct data type which has two members: location and length.

NSArray

An array is an ordered collection of things and if you're using NSArray, those "things" all have to be objects. Just like with an NSString, once it's created it cannot be modified unless you use NSMutableArray instead. Here's an example of making an array:

NSArray *myArray = [NSArray arrayWithObjects:@"Fish", @"Dogs", nil];

When creating an NSArray using the +arrayWithObjects: class method, the list of objects needs to be nil-terminated or it won't compile. It's not very intuitive at first, but it's important to remember to add it at the end of your list of objects or your code will crash and burn.

Organizing objects into an array is useless if you never do anything with your new collection. Here are some examples of what you can do with a filled array.

id myObject = [myArray objectAtIndex:3];
[myArray makeObjectsPerformSelector:@selector(runTowardsTheMoon)];
NSUInteger howMany = [myArray count];

The -objectAtIndex: instance method returns an id, a generic object in Cocoa. Since you can hold any type of object you want in an NSArray, when you access an object it gives you back a generic object type. It doesn't know what types of objects are being stored. Also, in the second example, what's a @selector? A @selector in Objective-C is basically a method signature. In this example, we're telling each object in the array to call their -runTowardsTheMoon instance method. In the third example the method returns an NSUInteger, a Foundation framework primitive type that's an unsigned integer, that is, an integer that's greater than zero.

NSDictionary

Almost all languages have the concept of a data structure that holds key-value pairs. Perl and Ruby have hashes, Python has dictionaries, PHP has associative arrays and Objective-C has the NSDictionary class. The Foundation framework provides us with NSDictionary (and its mutable brother NSMutableDictionary) to hold key-value pairs where all keys and values have to be objects. Let's define some dictionaries:

NSDictionary *myDict = [NSDictionary
  dictionaryWithObjectsAndKeys:@"aObj", @"aKey", @"bObj", @"bKey", nil];

NSArray *objs = [NSArray arrayWithObjects:@"One", @"Two", @"Three", nil];
NSArray *keys = [NSArray arrayWithObjects:@"Blue", @"Green", @"Yellow", nil];
NSDictionary *anotherDict = [NSDictionary
  dictionaryWithObjects:objs forKeys:keys];

In the first example we're calling the class method +dictionaryWithObjectsAndKeys: which has to be nil-terminated just like with some NSArray methods. In our object listing at the end we alternate object, key, object, key until we're done with all the pairs.

In the second example we're creating two different NSArray objects — one to hold the keys and another to hold the objects — and then we pass these two arrays into the NSDictionary class method +dictionaryWithObjects:forKeys: to build our dictionary. Again, notice that we're not using +alloc or -init to manually allocate memory for these dictionaries. This means we don't have to worry about their memory or calling -release when we're done using them.

A common use of NSDictionary is within an array: each element in the array is an NSDictionary object. For example, if you're building a Twitter app, the information for each tweet could sit in an NSDictionary, then each of these dictionaries is kept in order within an NSArray.

// First, find the dictionary in the 4th position in the array.
// Then, access the object paired with the key "tweet_text"
NSString *status = [[myArray objectAtIndex:4] objectForKey:@"tweet_text"];

Organizing NSDictionarys within an NSArray is a simple way to store some structured data.

NSNumber

If you want to store a primitive number in an NSArray or NSDictionary you'll have to package it up into an NSNumber object first. NSNumber has a number (ha!) of convenient class methods so you'll rarely need to manage the memory yourself.

NSNumber *myNumber = [NSNumber numberWithInt:326];
int myInt = [myNumber intValue];

You can get the primitive value of an NSNumber object just as easily as you can store it.

NSLog()

NSLog() is a C function to send output to the console, useful when debugging your application.

NSArray *myArray = [NSArray arrayWithObjects:@"Yo", @"Hey", nil];
NSLog( @"My array: %@", myArray );

int myInt = 2011;
int anotherInt = 2012;
NSLog( @"My int: %d and another int: %d", myInt, anotherInt );

Writing static text out to your log isn't that useful, so NSLog() lets you output variables as well. To include a variable in your output, you need to know what type of variable it is because NSLog() has different format specifiers based on the variable type. For example, an array is an object so you'd use %@. The specifier for a plain int is %d. These format specifiers are "stand-ins" for the actual value of the variable listed at the end of the function call. Apple provides the full list of string format specifiers to be used in NSLog() so make sure to always use the right one.

Building Cocoa Applications

We've talked about Objective-C syntax and what objects are. We've also discussed some common Foundation frameworks objects and how Objective-C classes are defined. Now it's time to start using some more complex Cocoa objects and putting all the pieces together to build a Cocoa app.

Models, Views and Controllers, Oh My!

A common design pattern in software development is called Model-View-Controller, or MVC for short. It's a methodology that separates the behavior and data of the application from the user interface. By keeping parts of the application separated into different classes your application can be developed more quickly and easily. By their design, the Cocoa frameworks practically force developers to use solid MVC principles, especially when building iOS apps.

Each object in your application is assigned one of three roles: model, view or controller. These roles describe that object's behavior and responsibilities. Let's talk about what each part of MVC means as it pertains to building iOS apps.

Model

The model layer manages the data of your application. Model objects encapsulate data and also hold the behaviors that will update or process this data. If you're building a Twitter app you might create model objects to represent an individual account, tweet, direct message or follower profile. If you were writing an invoice-sending app you'd probably have model objects for a company, person, invoice and more. A retro photography app? Perhaps only one model object representing a photo. You can think of model objects as being the nouns in a sentence that describes what your app does.

If you were building an app that persistently stores its data across launches, you'd build behavior into your models so that they could save their own data to a file on the user's iPhone, or perhaps send it to a server somewhere.

Here's some code showing how you'd initialize and use a model object.

Client *myClient = [[Client alloc] init];
myClient.firstName = @"John";
myClient.lastName = @"Smith";
myClient.phoneNumber = @"212-555-1212";
[myClient saveDataToFile];

In your apps you may use many model objects or you may use none, it all depends on what kind of app you're building. Model objects are typically just subclasses of the Cocoa root object, NSObject.

A benefit of using model objects is that they can be reused. Say you've built an iOS app and then you choose to port it to the Mac. If you've built your model objects with reusability in mind (not using iOS-specific APIs) it's fairly easy to port them to another Cocoa platform like OS X. Your model objects should not be concerned or tied to the user interface at all, thus making reusability possible.

View

Views are the objects used to build your user interface. For most of this guide the objects we've discussed represented abstract concepts, things or types of data, but view objects are different because they're actually seen by the user on the screen. Text labels, buttons, input fields, table views, scrollable panes and tabs are all view objects with defined behaviors and attributes. Views in iOS apps are all descendants of UIView, the root object used to build user interface components defined in the UIKit framework. Here's a list of all the classes available to you in UIKit.

Views have the ability to draw themselves (with colors, patterns, images, borders, shadows, corner radii, transparency, etc.) and also respond to user input. Apple provides a number of built-in view classes for use in your applications. You can use these as-is, customize some of their attributes, or go completely custom and write your own totally unique view objects to build custom user interfaces and new methods for user interaction.

Here's an example of initializing and customizing a UILabel, a built-in view object that draws a simple run of text on the screen.

CGRect textRect = CGRectMake(0,0,200,50);
UILabel *myLabel = [[UILabel alloc] initWithFrame:textRect];
myLabel.text = @"Photos";
myLabel.backgroundColor = [UIColor clearColor];
myLabel.font = [UIFont boldSystemFontOfSize:24];
myLabel.textColor = [UIColor blackColor];
myLabel.shadowColor = [UIColor whiteColor];

That's a lot of properties! You'll find that most Apple-built view objects come loaded with ways for them to be configured, mostly by setting their properties. Many properties on UILabel look similar to CSS styles like color, background color, text shadow, font and more.

An important thing to remember is that initializing a view object does not put it on the screen, it simply creates the object in memory. To put it on the screen, you must add the view as a subview on an existing user interface object. The main UIWindow for your application is the top-level view object for your application. All views that you create and use in your app are subviews of the window, or of each other, creating an overall view hierarchy.

Let's take a look at another view object example where we add it to the screen.

CGRect imageViewRect = CGRectMake(0,0,150,150);
UIImageView *logoView = [[UIImageView alloc] initWithFrame:imageViewRect];
logoView.image = [UIImage imageNamed:@"logo.png"];
[self.window addSubview:logoView];
[logoView release]; // Skip this if you're using ARC

When you initialize a view object you typically don't just call -init on it, you call a different initialization method -initWithFrame which not only initializes the view object but also sets its frame property. What's a frame? A frame is the rectangular region on the screen that the view is drawn into. It's a CGRect struct with four members: X position, Y position, width and height. The X and Y coordinates make up a CGPoint struct named origin, and the width and height are part of an CGSize struct named size.

In web design, when you absolutely position a <div> or any other block-level element, you define things like its position and dimensions. A view object is no different. Once a view object has been added to the overall view hierarchy (either as a subview of the window, or another view), it then has another property called its bounds. The bounds and frame properties are very similar in that they both describe the view's position and dimensions, but the bounds position is {0,0} whereas the frame position is relative to the parent view. If a 100x100px view object is placed 10px down and 10px from the top left of the parent view, its frame will be {10,10,100,100} and its bounds will be {0,0,100,100}. This distinction is important in a few places, mainly when you want to adjust the position of a view object once it's been placed. You'd do this by modifying its frame, not its bounds.

If you want to completely modify how a built-in interface widget is drawn, you'll want to create a subclass of it then re-implement its -drawRect: method and fill it with your own drawing code. Here's an example of how you might implement some custom drawing.

- (void)drawRect:(CGRect)rect {
	[[UIColor redColor] setFill];
	UIRectFill(self.bounds);

	[[UIColor whiteColor] setFill];
	CGRect thinLine = CGRectMake(0,0,self.bounds.size.width, 1);
	UIRectFill(thinLine);
	
	[[UIColor blueColor] setFill];
	NSString *springIsHere = @"Spring Is Here!";
	UIFont *springFont = [UIFont systemFontOfSize:24];
	[springIsHere drawAtPoint:CGPointMake(5,5) withFont:springFont];
}

Cocoa drawing follows what's called the "painter's model" for imaging. This means that each successive drawing operation is overlaid on top of the next so the order of your drawing code is very important. In this example we set the fill color to red and then fill the entire view's rectangle with that color. After this operation we set the fill color to white and fill a different rectangle with this new color. The CGRect that we define for the second drawing operation is a thin line that is 1px tall and spans the width of the entire view rectangle. It's drawn at {0,0} which is the top left of the view. Finally, we set the fill color to blue and draw a string at the point {5,5} with a 24px font.

Notice the call to self.bounds.size.width? If we take it step by step, we first make a call to self.bounds to access the bounds property on this UIView object. This returns a CGRect struct that, if you'll remember, has two members: origin and size. We then use dot notation to access the size member which is actually a struct again: an CGSize. Finally, we access the width member of this final CGSize struct to get the width of this view object's rectangle on the screen. Trust me, it's a lot more complex to describe it than to actually use it. You'll be accessing a view's size, position, X coordinate, width, size, etc., in this fast way all over your code.

This isn't a very exciting example of custom drawing, but in -drawRect: you could overlay graphics, vector shapes, colors and text any way you choose to implement a particular design. For interface widgets, a common pattern is to check what interaction state it's in before drawing to implement different designs for when a user's finger is tapping the control or other states.

Each view object takes up memory and if you have a very complex interface you may end up with many, many view objects on the screen and in memory all at once. This may be fine if you're presenting a fairly static-looking interface, but if you're building a complex scrollable container or table view, it just won't do at all. One way to get around a multitude of view objects on screen at once is to use only a few view objects (or just one!) and do all your text and shape drawing in -drawRect: as it's extremely fast. Instead of placing UIImageView objects into the view hierarchy, you'd instead draw a UIImage straight to the screen, and instead of using UILabel, you'd just draw strings directly at a CGPoint or within an CGRect.

Custom drawing in Cocoa is a very complex subject but it's an essential skill to have when building custom apps. Take a look at Apple's Introduction To Cocoa Drawing once you're ready to get dirty.

Controllers

Controllers are the meat of your application. They have many responsibilities and the bulk of your custom application code will be within controller classes. Here's a sampling of the responsibilities they have:

When developing iOS apps, controller objects are subclasses of the root controller class, UIViewController. Typically, a view controller will manage one screenful of data. If your iPhone app contains 3 screens of functionality, you'll have a UIViewController subclass managing each one of them. Take a look at the UIViewController class reference to see all the methods and properties that it provides.

As mentioned previously, when you write a subclass you're creating a specialized version of the parent class. By itself, UIViewController isn't very exciting. If you initialize a new UIViewController object (not your custom subclass) it won't work for you as it expects you to implement certain behaviors if you want things to happen. It's mainly a blueprint for you to use to build your own custom controller functionality.

UIViewController objects have a view property which is initialized as the top-level view object for a single screenful of content. Here's an example of creating this initial view object within the -loadView method and adding a UIImageView to it.

- (void)loadView {
	CGRect rect = [UIScreen mainScreen].applicationFrame;
	self.view = [[UIView alloc] initWithFrame:rect];
	
	UIImageView *imageView = [[UIImageView alloc] initWithFrame:CGRectMake(5,5,200,50)];
	imageView.image = [UIImage imageNamed:@"logo"];
	[self.view addSubview:imageView];
	[imageView release]; // Skip this if you're using ARC
}

The -loadView method in your UIViewController subclass should be implemented when you are programmatically creating your view hierarchy and not using Interface Builder. It's called automatically when you first access the view controller's view property (typically when you add it to the main window object) so there's no need to call it manually.

On the first line we're creating a CGRect to set as our main view's frame. The call to [UIScreen mainScreen]'s applicationFrame property returns a CGRect representing the usable section of the screen which is the entire thing, minus the 20px status bar at the top.

Next, we +alloc and -init our UIView object and set it to self.view which is this controller's view property. Now that our top-level view object has been initialized, we add a UIImageView to it as a subview, then release that view.

Let's now add a UIButton object to our view and go over how we handle tap events.

UIButton *myButton = [UIButton buttonWithType:UIButtonTypeInfoLight];
[myButton addTarget:self action:@selector(wasTapped:) forControlEvents:UIControlEventTouchUpInside];
[myButton setFrame:CGRectMake(20,20,24,24)];
[self.view addSubview:myButton];

Here we're initializing a UIButton using the convenient class method +buttonWithType and setting its type to UIButtonTypeInfoLight which is a small info button built into UIKit. Next, we're calling -addTarget:action:forControlEvents: which is the standard way to setup a method getting called when a user interaction occurs. It's an instance method on the UIControl class, a parent class of UIButton. It takes in three arguments: which class will handle the action (in this case, self), which @selector will be called, and finally, which interaction event we want to pay attention to. In this example we're listening for UIControlEventTouchUpInside which is basically a normal tap on an iPhone or iPad screen. All these event types are listed in the UIControl class reference. All that's left is to implement the method -wasTapped: in this class and you're in business. Lastly, we set the button's frame and then add it as a subview on the main view object.

So where do we initialize our view controller and how do we begin using it? Well, if this is the view controller that will show the main screen of your app, it will be initialized in the application delegate class which Xcode creates for you when you create a new iOS project.

- (BOOL)application:(UIApplication *)application didFinishLaunchingWithOptions:(NSDictionary *)launchOptions {    	
	self.myViewController = [[MyViewController alloc] initWithNibName:nil bundle:nil];
	[self.window addSubview:self.myViewController.view];
	[self.myViewController.view release]; // Skip this if you're using ARC
	
	[self.window makeKeyAndVisible];
}

In our application delegate class, the instance method -application:
didFinishLaunchingWithOptions:
gets called when the app has successfully finished launching and is ready to be configured. Here we set self.myViewController (a property you'd create on this class) to the newly-created MyViewController instance, our subclass of UIViewController. Next, we add its view as a subview on the main window. It's at this exact moment that the runtime automatically calls -loadView on the view controller and our custom view creation code is executed.

I've only scratched the surface of how controller objects are used. For a more in-depth look, read Apple's View Controller Programming Guide.

Your First App

If you haven't already done so, open up Xcode and create a new iOS application. It will prepare a project for you with some key files already created. One of them is your application's delegate class which means it handles the main setup of your application. In this class you'll find -application:didFinishLaunchingWithOptions: which is the jump-off point for your entire app. Here is where your first code will go, where your first UIViewController classes will be initialized, where your first model objects may be created and used.

If you want to get up and running fast with a detailed, step-by-step tutorial, head to the main Design Then Code site and take a look at the first tutorial for an app called Texture! It's over 70 pages of text, screenshots and code snippets, including the full Photoshop file and Xcode project.

-Mike